Background: Frith and Happe (1994) describe central coherence (CC) as the normal tendency to integrate individual elements into a coherent whole, a cognitive style which varies in the general population. Individuals with autism are at the extreme (weak) end of the continuum of coherence. There has been debate over whether CC is independent from other psychological functions such as theory of mind and executive control. Methods: We examined the validity of the CC construct in 76 typically developing 4- and 5-year-old children using four visuospatial coherence tasks. We also investigated the extent to which individual differences in performance on CC measures were associated with individual differences on measures of mindreading and executive functioning (EF). Results: Unexpectedly, the CC variables were not highly intercorrelated and did not form a single coherence factor. Instead, the results revealed a two-factor structure, one of which corresponded to visuospatial constructional ability. Correlations between the emerging factors of coherence and mindreading ability were weak, even when developmental differences in age and verbal and nonverbal ability were taken into account. In contrast, visuospatial constructional ability was reliably related to measures of EF, whilst correlations between EF and the second CC factor were very weak. Conclusions: CC in typically developing preschoolers is not a unitary construct. Whilst there is little relationship between CC and mindreading abilities in these young children, executive control appears to be associated with one aspect of coherence: visuospatial construction.